As part of the respiratory system, human lungs are responsible for exchanging air with the blood to facilitate human breathing. During the process of oxygen and carbon dioxide exchange, the lungs are exposed to a risky environment and can suffer from diseases such as cancer, asthma, and influenza. To study lung conditions and respiratory diseases, preclinical models are often used, although they can be accompanied by many limitations. In response to these challenges, lung-on-a-chip (LOAC) devices are a promising new approach. The lung is a complex organ with tree-like airways extending from the trachea and ending with millions of alveoli where air exchange occurs. To mimic the in vivo microarchitecture and mechanical conditions of the lung, tissue-tissue interactions, air-liquid interfaces and mechanical stretching due to breathing must be reconstituted (De et al., 2022).

Huh et al. (2010) implemented the first LOAC using a human lung alveolar epithelial cell line. This development successfully replicated organ-level physiological and pathophysiological responses on a chip. As shown in Figure 1A, the lung chip consists of two parallel microchannels separated by a thin polydimethylsiloxane (PDMS) porous membrane coated with extracellular matrix (ECM). Human alveolar epithelial cells were cultured in the upper air channel, where air could be introduced, while human pulmonary microvascular endothelial cells were cultured in the bottom blood channel to achieve an alveolar-capillary interface. Two chambers beside the double-layer culture area were designed to produce elastic deformation of the microchannels, thus performing mechanical stretching in the microsystem to replicate breathing movements.

An in silico model of Huh et al.’s microdevice was developed by Hancock and Elabbasi (2015) using COMSOL Multiphysics^®. The model configured the architecture, materials, laminar fluid flow and applied mechanical forces within the chip based on parameters from previous literature (Huh et al., 2010) (see the schematic shown in Figure 1A) and analysed the fluidic dynamics across the device. Computational simulations provided estimates of the transport of diluted species, such as drugs and nutrients, across the membrane and computed the shear stress on cells. Subsequently, Hancock and Elabbasi (2016) further advanced this work by introducing an effective elastic modulus of the porous membrane into the previous model. The enhanced model could be used to compute the effective molecular diffusivity and effective thermal diffusivity near the porous membrane. These factors influence heat and mass transfer through the air-liquid interface of the device, and their impact on cell cultures such as shear stress and biochemical gradients are worthy of further study. However, neither model explicitly includes the multicellular structure that represents the organ in the LOAC device, and so its growth and development, and how that may depend on the device design must still be studied experimentally.

One application of the LOAC is to model the transport of nanoparticles (NPs) in the lung. Nanoparticles range in size from 1 to 100 nm and can originate from sources such as cigarette smoke, air pollution, or viruses. In the post-COVID-19 pandemic period, respiratory-related diseases have become a concern, and clinical research on NPs has attracted increased attention. In the pursuit of more effective testing of NPs, Amin Arefi et al. (2020) proposed an in silico model to study the transport and absorption of NPs within a LOAC. The simulation was based on the design of Huh et al.‘s LOAC (Figure 1A) and aimed to estimate NP transport under various breathing patterns and to investigate the potential effects that different NP sizes may introduce. Using COMSOL to solve the Navier-Stokes equations for fluid flow in the chip, and employing Eulerian advection-diffusion, Lagrangian particle tracking for fine and coarse NP tracking, and Langmuir and Frumkin kinetics for computing the surface adsorption and desorption, the results indicated that particle deposition increases during breathing with high frequency or long breath-holding time. Additionally, the findings highlighted that both particle sizes and the orientation of the lung epithelium layer relative to gravity affect deposition efficiency.

The human intestine is responsible for a range of digestion, absorption, and secretion processes. Serving as a barrier between the intestinal contents and other parts of the human body, it plays a crucial role in oral drug absorption. Meanwhile, physiologically, the intestine hosts a rich microbiome and communicates with other organs to regulate body function. To model these properties, dynamic mechanical stress, precise signal transport, and co-culture ability are required to be replicated in a model culture system (Bein et al., 2018).

To date, one of the most commonly used devices in experimental work related to gut-on-a-chip (GOAC) is the chip designed by Emulate Inc., initially developed and published by Huh et al. (2010). Kim et al. (2012) first applied the microfluidic chip technique for GOAC, subesequently developed by Kim and Ingber (2013). With a design similar to the device described earlier in section 2.1 (Figure 1A), this chip also features two parallel channels separated by an ECM-coated porous membrane, allowing observation of diffusion between two channels. PDMS was used as the material for both the chip and membrane. In this configuration, human Caco-2 intestinal epithelial cells (Caco-2 human colorectal carcinoma line (Peterson and Mooseker, 1992)) were cultured in the upper channel, while the culture medium was perfused through both channels at the same flow rate. The epithelial monolayer forms on the porous membrane, and the peristaltic motions of the intestine epithelium are mimicked by the tractions applied by the two side chambers to the membrane in between. Specifically, the formation of finger-like structures called intestinal villi, lined by polarized epithelial cells and separated by crypts, was observed. Furthermore, Kasendra et al. (2018) presented additional studies of GOAC using human intestinal epithelial cells derived from organoids and microvascular endothelial cells. In this work, organoid cells grew into epithelium with finger-like structures, closely mimicking the structure of the human duodenum in vivo and also performing the intestinal barrier functions.

Komen et al. (2020) designed a different GOAC with the specific aim of exposing cancer cells to the drug oxaliplatin through dynamic perfusion. The chip design is illustrated in Figure 1B, featuring two crossing channels. The upper channel is for drug dosing and the sealed lower channel is for seeding cells. The cell culture takes place in the intersection of the two channels. In silico modelling with COMSOL Multiphysics^® in two space dimensions was used to predict the drug concentration for the lower channel over time. Within the simulation, the two-dimensional diffusion equation was solved for the flux and concentration of the solute according to Fick’s law. An optimal value of membrane porosity was calculated and used in subsequent experiments to design the thickness and pore size of the membrane.

The liver is a vital metabolic organ with many essential functions, including protein synthesis and organism detoxification. A liver chip is typically expected to replicate the architectural features, metabolic capabilities, regenerative properties, and injury response of a living liver. Within the liver, the liver sinusoids play a crucial role in regulating blood flow and thus become a key component to be reconstructed on a liver chip. A typical liver sinusoidal chip may consist of four main types of hepatic cells: liver sinusoidal endothelial cells, Kupffer cells, hepatic stellate cells, and hepatocytes within two fluid channels, i.e., sinusoidal vascular channel and Disse space (Deguchi and Takayama, 2022).

A microfluidic system mimicking the liver sinusoidal structure was developed by Du et al. (2017) (Figure 1C). Similar to the chip designed by Huh et al. (2010) (Figure 1A), the liver chip consists of two PDMS layers separated by a porous polyester membrane, with Collagen I pre-coated on both the channels and the membrane. The Kupffer cells and liver sinusoidal endothelial cells are cultured on the membrane, while the hepatic stellate cells and hepatocytes are cultured in the lower channel. As a result, the fundamental architecture of liver sinusoids was replicated, an in vivo monolayer of liver sinusoidal endothelial cells was successfully established, and metabolic activity was activated through the co-culture of hepatocytes and nonparenchymal cells. Du et al. (2017) also constructed an in silico model: a computational fluid dynamics simulation of the chip geometry was constructed in FLUENT software, and particle tracking visualization tests were used to estimate the flow field inside the chip at different heights. The computational results were validated against experimental data. By configuring the parameters of the chip geometry and considering the velocity of the laminar flow through the system and viscosity of the fluid, the fluid flow and mass transfer inside the chip could be better understood.

Another advanced liver chip design is the very large-scale liver-lobule (VLSLL)-on-a-chip proposed by Banaeiyan et al. (2017) (Figure 1D). The two-layer PDMS device consists of multiple hexagonal culture chambers that mimic the natural geometry of liver lobule: a combination of the hepatic lobule structure and the blood-circulation network. Each chamber has a bottom layer for tissue culture and a top layer for the seed-feed network, connected via a central opening that mimics the central vein. This design helps protect the cultured cells from the high shear stress of fluid flow. Unlike typical designs, this system relies on diffusion-convection effect for nutrient delivery, and it is validated through in silico models. Fluid dynamics within the device were analyzed to optimize key design parameters, such as flow velocity and shear stress. The model also explored glucose diffusion across the chambers, confirming that diffusion continues until the glucose concentration in the center of the chamber reaches a steady state. These simulations were crucial for predicting nutrient transport within the chip and optimization of the device. As a result, the chip provides a robust platform for studying liver function and high-throughput drug metabolism. These functionalities were experimentally validated by culturing human hepatocellular carcinoma cells and human-induced pluripotent stem cell-derived hepatocytes separately on the proposed device.

Subsequent work by Chen et al. (2020), based on (Du et al., 2017), further demonstrated that in silico flow field simulation could enhance chip design. An existing problem in the chip design is that chip geometry parameters such as pore size, channel size, membrane thickness, and membrane permeability all have a significant influence on the flow features inside the device and can affect the culture functions. A systematic analysis is required in this case to quantify their relationship and optimize the design. Chen et al. (2020) used immersed boundary methods to construct the in silico model framework in MATLAB, with theoretical modelling supporting assumptions of the immersed boundary method model. The simulations investigated the dependence of fluid dynamics on chip geometry (height, width), membrane permeability, and fluid viscosity. The results showed that increasing the lower channel height initially increased and then decreased the wall shear stress. It was also found that increasing the channel length or membrane permeability enlarged the transition threshold of the maximum wall shear stress. Moreover, the transmembrane pressure difference and wall shear stress increased with increasing viscosity. Chen et al. (2020) suggest that this may indicate hepatocytes could experience higher wall shear stress as blood viscosity increases, although their model does not explicitly include the cellular structures typically found in the liver.

In silico models have also been used to aid in predicting the oxygen level in a liver chip. To understand cell culture with hypoxic or anoxic conditions, and examine cell viability and behaviour under specific oxygen-levels, thermoplastic devices made from oxygen-impermeable materials are considered as an alternative to PDMS, Ochs et al. (2014) developed a computational model to predict oxygen levels in devices made of oxygen-permeable materials PDMS and poly (methyl pentene), and the oxygen-impermeable material cyclic olefin copolymer during cell culture. Assuming that endothelial cells and hepatocytes were cultured on the chip with a known oxygen uptake rate, Ochs et al. (2014) estimated cell respiration in devices; computational results were validated experimentally. Their results verified that moderate media flow rates or an oxygen-permeable film can help counteract oxygen depletion in non-permeable devices. Finally, poly (methyl pentene) was introduced as a suitable material due to its good oxygen permeability and property of maintaining sufficient oxygen levels for high metabolism cell types.